From 5e286478f7f3be9bd996be678ba269030de5ae72 Mon Sep 17 00:00:00 2001
From: Grant Lemons <grantlemons@aol.com>
Date: Thu, 16 Jan 2025 09:17:09 -0700
Subject: [PATCH] fix(title-case): retain all characters from input
---
harper-core/src/title_case.rs | 34 +++++++++++++++++++++++++++-------
1 file changed, 27 insertions(+), 7 deletions(-)
diff --git a/harper-core/src/title_case.rs b/harper-core/src/title_case.rs
index 7485017c..cad47f6c 100644
--- a/harper-core/src/title_case.rs
+++ b/harper-core/src/title_case.rs
@@ -28,10 +28,8 @@ pub fn make_title_case(toks: &[Token], source: &[char], dict: &impl Dictionary)
return Vec::new();
}
- let start_index = toks.first().unwrap().span.start;
-
let mut words = toks.iter_word_likes().enumerate().peekable();
- let mut output = toks.span().unwrap().get_content(source).to_vec();
+ let mut output = source.to_vec();
// Only specific conjunctions are not capitalized.
lazy_static! {
@@ -62,17 +60,16 @@ pub fn make_title_case(toks: &[Token], source: &[char], dict: &impl Dictionary)
|| words.peek().is_none();
if should_capitalize {
- output[word.span.start - start_index] =
- output[word.span.start - start_index].to_ascii_uppercase();
+ output[word.span.start] = output[word.span.start].to_ascii_uppercase();
// The rest of the word should be lowercase.
- for v in &mut output[word.span.start + 1 - start_index..word.span.end - start_index] {
+ for v in &mut output[word.span.start + 1..word.span.end] {
*v = v.to_ascii_lowercase();
}
} else {
// The whole word should be lowercase.
for i in word.span {
- output[i - start_index] = output[i].to_ascii_lowercase();
+ output[i] = output[i].to_ascii_lowercase();
}
}
}
@@ -120,6 +117,29 @@ mod tests {
)
}
+ #[test]
+ fn span_breaks_in_source() {
+ use super::make_title_case;
+ use crate::{CharStringExt, Span, Token, TokenKind, WordMetadata};
+ use itertools::Itertools;
+
+ let source = "// linted
+// linted";
+ let source = source.chars().collect_vec();
+
+ let toks = &[
+ Token::new(Span::new(3, 9), TokenKind::Word(WordMetadata::default())),
+ Token::new(Span::new(9, 10), TokenKind::Newline(1)),
+ Token::new(Span::new(13, 19), TokenKind::Word(WordMetadata::default())),
+ ];
+
+ assert_eq!(
+ make_title_case(toks, &source, &FstDictionary::curated()).to_string(),
+ "// Linted
+// Linted"
+ )
+ }
+
#[derive(Debug, Clone)]
struct Word(String);